Dynamic Sound Identification

Dynamic sound recognition is also known as “query-by-example,” it was recently found commercial success as a means to identify music through short audio snippets, captured through a microphone. The process behind this is that the first-generation algorithms recognized unique signatures in a particular sound, which they could then match with a most likely source or an equivalent sound store in a large database of previously identified auditory signature. It helps the music listeners to identify a sing’s title and the performing lists through use of algorithms. The most important output of this research was a mobile app, dubbed Epimetheus. It was proficient at recognizing music, advertisements and human voices. Epimetheus differed compared to other similar apps by being able to pick up subtle auditory signals and sorting through environmental noise in order to accurately identify natural phenomena, such as the changing tides. Scientific researchers meant to benefit from this function, by employing Epimetheus as a tool to track ecological change in remote locations, by incorporating a machine learning algorithm that adapts to new inputs and provides users with useful information about the sounds that are being processed. The app was popular between students and hobbyists. The app could identity personal information about those who are speaking, links to websites selling a product being advertised on television, encyclopedic entries about bird call in the wild and other relevant resources.

Example:

A problem arose when the app was in the process of testing. There was a transgender by the name of Sybel, she was born as a male and now identifies as a female. When she tested her voice on the system, based on her voice sample Epimetheus identified Sybel as a male and showed further information about her history, including a link to several online videos that showed Sybel prior to her transition. From this testing the researchers realized that it the potential harm it could cause to transgender individuals.

 

https://aiethics.princeton.edu/wp-content/uploads/sites/587/2018/10/Princeton-AI-Ethics-Case-Study-2.pdf